The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs various automated shell operations including git commit, git revert, and grep to manage version control and analyze local skill files. It also utilizes npx to execute external tools such as playwright.
[COMMAND_EXECUTION]: The script scripts/screenshot.mjs uses the child_process.execSync method to run the open command on the host operating system, which is used to automatically display generated report images.
[REMOTE_CODE_EXECUTION]: The script scripts/screenshot.mjs contains a hardcoded absolute path to a Node.js module (/Users/alchain/.npm-global/lib/node_modules/playwright/node_modules/playwright-core). Relying on hardcoded local paths for module resolution is a security anti-pattern that creates environment-specific dependencies.
[EXTERNAL_DOWNLOADS]: The documentation suggests downloading a zip archive of the skill from an external Cloudflare R2 bucket (pub-161ae4b5ed0644c4a43b5c6412287e03.r2.dev), which is not an officially verified or trusted source.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from other SKILL.md files on the user's filesystem.
Ingestion points: Reads SKILL.md files from the .claude/skills/ directory during Phase 0 and Phase 1.
Boundary markers: The skill logic does not implement boundary delimiters or 'ignore' instructions when reading the content of the skills being optimized.
Capability inventory: The skill has the ability to execute shell commands (git, grep, node, npx) and perform file-write operations.
Sanitization: There is no evidence of sanitization or filtering of the external skill content before it is processed by the agent.

darwin-skill